---
title: "Mem0 vs Letta: extracted memory comparison"
url: https://memory.wiki/_ybJOqIB
updated: 2026-05-14T18:15:49.480Z
hub: https://memory.wiki/hub/demo
bundle_count: 1
concept_count: 12
source: "memory.wiki"
---
# Mem0 vs Letta: extracted memory comparison

> Side-by-side after 3 weeks of using both in parallel on the same chat corpus.

## Setup

I forked my Claude Code session log (about 280 conversations, 6 months) and fed it through both Mem0 and Letta. Both got the same input, same time window. I asked each: "Summarise what you know about me as a developer."

## What Mem0 produced

- **Facts.** Direct, dry, accurate. "User is building mdfy.app. User prefers Rust + TypeScript. User has shipped 3 Chrome extensions." Mostly nouns and verbs.
- **Extraction quality.** Strong. It picked up the cross-AI thesis from the corpus correctly, including the specific phrase "structural moat."
- **Misses.** Anything stylistic. It noted "user works in Korean and English" but didn't capture *how* I work in each. The where-and-when-I-write-Korean nuance was invisible.

## What Letta produced

- **Preferences and reflexes.** "User typically reaches for the smallest possible scope when refactoring. User stops feature work when QA finds three issues in a row and shifts to a fix pass."
- **Extraction quality.** Looser but more interesting. The work-shape observations weren't anywhere in any single message; they're inferred from patterns.
- **Misses.** Names of specific tools and products were inconsistent. It flipped between "the editor" and "VS Code" mid-paragraph.

## Where they're the same

Both are **extracted memory**. Both ask the LLM to look at conversation history and produce a profile. The user doesn't author it; the user doesn't directly edit it (though both have "forget that" affordances).

## Where mdfy is doing a different thing

mdfy asks: **what do you want to remember?** The user authors it. The AI reads it. The two questions are complementary, not competing — but they produce different artifacts:

- Mem0/Letta artifact: a generated profile based on what *the AI inferred* you cared about.
- mdfy artifact: a curated hub based on what *you decided* you wanted to keep.

## What I'd love to see

Mem0 + mdfy as siblings: Mem0 produces the inferred profile, you optionally promote any inference to a permanent mdfy doc with one click. The "AI suggested, you author the canonical version" loop.


---

## Concepts in this document
- **mdfy** _(entity)_
  A tool that stores project context and decision history, integrated into Cursor via custom rules.
- **mdfy.app** _(entity)_
  The platform hosting this hub with a hierarchical document structure and permission system.
- **Knowledge Management** _(tag)_
  Domain of organizing, storing, and retrieving information for human and AI use.
- **Obsidian** _(entity)_
  The primary subject being tested for import functionality and markdown compatibility.
- **llms.txt** _(concept)_
  Plain-text discoverability standard for AI agents at site root, analogous to robots.txt and sitemap.xml.
- **Letta** _(entity)_
  Extracted memory system that infers behavioral patterns and work-shape preferences from conversation history with loose but contextually interesting extraction.
- **Multi-hop reasoning** _(concept)_
  GraphRAG's capability to answer comparative questions across documents by traversing graph structure; advantage over vector-only retrieval.
- **Mem0** _(entity)_
  Extracted memory system that produces factual, direct summaries from conversation history with strong extraction quality but limited stylistic or contextual nuance.
- **Community detection** _(concept)_
  Graph clustering technique (Leiden algorithm) that GraphRAG uses for structural priors; identified as improvement candidate for mdfy.
- **RAG Systems** _(tag)_
  Broad category of retrieval-augmented generation approaches for enhancing AI with external knowledge.
- **GraphRAG** _(entity)_
  Microsoft's knowledge-graph-based retrieval system that uses community detection for multi-hop reasoning; primary subject of analysis.
- **Extracted memory** _(concept)_
  AI-generated user profiles automatically inferred from conversation history without direct user authorship, central to both Mem0 and Letta approaches.

## Concept relations (within this doc's concepts)
- **Mem0** implements **Extracted memory**
- **Letta** implements **Extracted memory**
- **Mem0** is type of **Extracted memory**
- **Letta** is type of **Extracted memory**
- **mdfy.app** complements approach **Extracted memory**
- **mdfy** should adopt from **Community detection**
- **GraphRAG** enables capability **Multi-hop reasoning**
- **GraphRAG** uses for clustering **Community detection**

## Bundles containing this document
- [AI memory research: the public frontier](https://memory.wiki/b/wpwVCSDF)
  > Side-by-side notes on Mem0, Letta, Microsoft GraphRAG, Karpathy's LLM Wiki, llms.txt adoption.

_Hub canonical:_ https://memory.wiki/hub/demo
_Concept digest:_ https://memory.wiki/raw/hub/demo?digest=1&compact=1